block structure summary /rotation matrix

[quantumable]

Dear All,
I have some concerns when running my AFM calculation with soliDMFT
-1) block structure summary: I have a system also with two impurities like the NdNiO2 example I used the same type of input file but the block structure summary is written differently
for my calcul I have this the output : block structure summary

-- Shell type #  0: [0]
 | shell multiplicity 1
 | block struct. sum_k : [('up', 5), ('down', 5)]
 | block struct. solver: {'down_0': 4, 'down_1': 1, 'up_0': 4, 'up_1': 1}
 | deg. orbitals : []
-- Shell type #  1: [1]
 | shell multiplicity 1
 | block struct. sum_k : [('up', 5), ('down', 5)]
 | block struct. solver: {'down_0': 2, 'down_1': 1, 'down_2': 1, 'down_3': 1, 'up_0': 2, 'up_1': 1, 'up_2': 1, 'up_3': 1}
 | deg. orbitals : []

for NdNiO2 example I have this in output :

block structure summary
 -- Shell type #  0: [0]
  | shell multiplicity 1
  | block struct. sum_k : [('up', 5), ('down', 5)]
  | block struct. solver: {'down_0': 1, 'down_1': 1, 'down_2': 1, 'down_3': 1, 'down_4': 1, 'up_0': 1, 'up_1': 1, 'up_2': 1, 'up_3': 1, 'up_4': 1}
  | deg. orbitals : [['up_1', 'up_3'], ['up_2', 'down_0'], ['down_1', 'down_3']]
 -- Shell type #  1: [1]
  | shell multiplicity 1
  | block struct. sum_k : [('up', 5), ('down', 5)]
  | block struct. solver: {'down_0': 1, 'down_1': 1, 'down_2': 1, 'down_3': 1, 'down_4': 1, 'up_0': 1, 'up_1': 1, 'up_2': 1, 'up_3': 1, 'up_4': 1}
  | deg. orbitals : [['up_1', 'up_3'], ['up_2', 'down_0'], ['down_1', 'down_3']]

and respectively

h5ls 2d.h5/DMFT_results/last_iter/G0_freq_0
block_names              Group
down_0                       Group
down_1                        Group
up_0                            Group
up_1                             Group

h5ls nno.h5/DMFT_results/last_iter/G0_freq_0
block_names              Group
down_0                   Group
down_1                   Group
down_2                   Group
down_3                   Group
down_4                   Group
up_0                     Group
up_1                     Group
up_2                     Group
up_3                     Group
up_4                     Group

why is the writing different?

2-)here are the two images representative of my system in front view and in profile view. the system constitutes two molecule rotate of 45 degree one of the other. since I wish to analyze the different orbital of d. do I have to rotate the matrix of rotation with the parameter TRANSFORM(pro.cfg) ? if yes how to write that matrix.

thank you

Bests,

[quantumable]

by doing the messages written in the output files of the calculation of my system and of your example NdNiO2 I have this

Mine:

 *** determination of block structure ***
magnetic calculation: removing the spin degeneracy from the block structure
using 1-particle density matrix and Hloc (atomic levels) to determine the block structure

 number of ineq. correlated shells: 2

 block structure summary
 -- Shell type #  0: [0]
  | shell multiplicity 1
  | block struct. sum_k : [('up', 5), ('down', 5)]
  | block struct. solver: {'up_0': 4, 'up_1': 1, 'down_0': 4, 'down_1': 1}
  | deg. orbitals : []
 -- Shell type #  1: [1]
  | shell multiplicity 1
  | block struct. sum_k : [('up', 5), ('down', 5)]
  | block struct. solver: {'up_0': 2, 'up_1': 1, 'up_2': 1, 'up_3': 1, 'down_0': 2, 'down_1': 1, 'down_2': 1, 'down_3': 1}
  | deg. orbitals : []

NdNiO2 exemple

Warning: Imaginary part in density will be ignored (7.714583614885123e-14)
    -0.275714 < Chemical Potential < -0.003622
    41.193241 < Total Density < 42.000548
    Chemical Potential found in 4 iterations :
    Total Density = 42.000548;Chemical Potential = -0.003622

 number of ineq. correlated shells: 2

 block structure summary
 -- Shell type #  0: [0]
  | shell multiplicity 1
  | block struct. sum_k : [('up', 5), ('down', 5)]
  | block struct. solver: {'down_0': 1, 'down_1': 1, 'down_2': 1, 'down_3': 1, 'down_4': 1, 'up_0': 1, 'up_1': 1, 'up_2': 1, 'up_3': 1, 'up_4': 1}
  | deg. orbitals : [['up_1', 'up_3'], ['up_2', 'down_0'], ['down_1', 'down_3']]
 -- Shell type #  1: [1]
  | shell multiplicity 1
  | block struct. sum_k : [('up', 5), ('down', 5)]
  | block struct. solver: {'down_0': 1, 'down_1': 1, 'down_2': 1, 'down_3': 1, 'down_4': 1, 'up_0': 1, 'up_1': 1, 'up_2': 1, 'up_3': 1, 'up_4': 1}
  | deg. orbitals : [['up_1', 'up_3'], ['up_2', 'down_0'], ['down_1', 'down_3']]

I have this message:

*** determination of block structure ***
magnetic calculation: removing the spin degeneracy from the block structure
using 1-particle density matrix and Hloc (atomic levels) to determine the block structure

also and additional block : DFT density matrix

I joint here the two output files

Output_Mine.zip
Output_NdNiO2.zip

[the-hampel]

Hi,
what exactly is your concern? In both cases you start with 5 orbitals on the lattice Gf level:

| block struct. sum_k : [('up', 5), ('down', 5)]

And in your case this is converted for the solver to a 4x4 full matrix form and a separate 1x1 block for each spin channel:

| block struct. solver: {'down_0': 4, 'down_1': 1, 'up_0': 4, 'up_1': 1}

This is correct if you expect 1 orbital to be decoupled from the other orbitals. You have to judge if this is correct by the symmetry of your system. For example in the case of the NdNiO2 you find a diagonal form with five 1x1 blocks for each spin channel:

| block struct. solver: {'down_0': 1, 'down_1': 1, 'down_2': 1, 'down_3': 1, 'down_4': 1, 'up_0': 1, 'up_1': 1, 'up_2': 1, 'up_3': 1, 'up_4': 1}

meaning that the local problem and the hybridization is completely diagonal. This is in most cases an approximation, but should hold up to the set threshold of block_threshold in the dmft_config.ini file.

I don't understand what you mean with "writing is different"? Please check the symmetry of your system and make sure this is what you expect. Generally it should be relatively safe to trust the outcome of the block structure finder.

Best,
Alex

[quantumable]

Hi,

Indeed I would like to know if something was wrong in my calculation. because I have the different writing of the "block structure summary".

for this input,

[general]
seedname = Mn
jobname = NNO_AFM
block_threshold = 0.01
solver_type = cthyb
n_iw = 8001
n_tau = 50001
prec_mu = 0.001
h_int_type = density_density
U = 4.0
J = 1.0
beta = 194
magnetic = True
magmom = -0.01, 0.01
afm_order = True

n_iter_dmft = 12

g0_mix = 0.6

dc_type = 0
dc = True
dc_dmft = False
[solver]
length_cycle = 150
n_warmup_cycles = 10000
n_cycles_tot = 2e+8
imag_threshold = 1e-11
n_cores = 96
perform_tail_fit = True
fit_max_moment = 6
fit_min_w = 8
fit_max_w = 14
measure_density_matrix = True

I have this:

block structure summar
 -- Shell type #  0: [0]
  | shell multiplicity 1
  | block struct. sum_k : [('up', 5), ('down', 5)]
  | block struct. solver: {'up_0': 1, 'up_1': 1, 'up_2': 1, 'up_3': 1, 'up_4': 1, 'down_0': 1, 'down_1': 1, 'down_2': 1, 'down_3': 1, 'down_4': 1}
  | deg. orbitals : [['up_1', 'up_3'], ['up_2', 'down_1', 'down_3']]
 -- Shell type #  1: [1]
  | shell multiplicity 1
  | block struct. sum_k : [('up', 5), ('down', 5)]
  | block struct. solver: {'up_0': 1, 'up_1': 1, 'up_2': 1, 'up_3': 1, 'up_4': 1, 'down_0': 1, 'down_1': 1, 'down_2': 1, 'down_3': 1, 'down_4': 1}
  | deg. orbitals : [['up_1', 'up_3'], ['up_2', 'down_1', 'down_3']]

And for this,

general]
seedname = Mn
jobname = NNO_AFM

block_threshold = 0.0001

solver_type = cthyb
n_iw = 8001
n_tau = 50001

prec_mu = 0.001

h_int_type = density_density
U = 4.0
J = 1.0

beta = 194

magnetic = True
magmom = -0.5, 0.5
afm_order = True

n_iter_dmft = 12

g0_mix = 0.6

dc_type = 0
dc = True
dc_dmft = False

[solver]
length_cycle = 150
n_warmup_cycles = 10000
n_cycles_tot = 2e+8

imag_threshold = 1e-5
n_cores = 96
perform_tail_fit = True
fit_max_moment = 4
fit_min_w = 4
fit_max_w = 14
measure_density_matrix = True

I got this:

 block structure summary
 -- Shell type #  0: [0]
  | shell multiplicity 1
  | block struct. sum_k : [('up', 5), ('down', 5)]
  | block struct. solver: {'up_0': 5, 'down_0': 5}
  | deg. orbitals : []
 -- Shell type #  1: [1]
  | shell multiplicity 1
  | block struct. sum_k : [('up', 5), ('down', 5)]
  | block struct. solver: {'up_0': 5, 'down_0': 5}
  | deg. orbitals : []

and I don't understand

[the-hampel]

That makes totally sense. For the second calculation you decreased block_threshold from 0.0001 to 0.01 , and your resulting block structure became more diagonal, i.e. having more blocks with smaller size. Does this not make sense to you?